search

sanitizers / patchback-github-app

https://github.com/apps/patchback
GNU General Public License v3.0
12 stars 4 forks source link

Fix remote add command comment #22

Closed AustinScola closed 3 years ago

AustinScola commented 3 years ago

Fix the comment for adding a remote repository.

AustinScola commented 3 years ago

Address Issue #21

webknjaz commented 3 years ago

Thanks!

AustinScola commented 3 years ago

No problem! This is a cool GitHub App that you have created. Out of curiosity, what percentage of the time have you observed that the patchbacks apply cleanly vs. require manual conflict resolution?

webknjaz commented 3 years ago

The idea actually comes from what CPython folks did for their project (that is a user account-based bot, not a GitHub App, though).

I don't have the stats. I think @ansible-collections is the biggest consumer, maybe @felixfontein could answer the question about the clean cherry-picks.

In general, this depends on how different the target branch is compared to the main are + the order of backporting the subsequent PRs that edit the same file.

AustinScola commented 3 years ago

The idea actually comes from what CPython folks did for their project (that is a user account-based bot, not a GitHub App, though).

Ahh, cool. I wonder if that was made before GitHub Apps were around. I also saw this is using gidgethub which I think was spawned out of CPython as well.

In general, this depends on how different the target branch is compared to the main are + the order of backporting the subsequent PRs that edit the same file.

Interesting, I hadn't considered the case that the ordering might not be the same.

felixfontein commented 3 years ago

I don't have the stats. I think @ansible-collections is the biggest consumer, maybe @felixfontein could answer the question about the clean cherry-picks.

It depends a lot on the repository I'd say, and how much activity happens to parts of it, and how much of that activity is backported and how much not (breaking changes vs. features vs. bugfixes). In case of community.general, I'd say that it works most of the time. There are some situations where I'm pretty sure it probably won't work and I'm trying it anyway, to safe some manual work in case it does work (sometimes git is more clever than expected).

AustinScola commented 3 years ago

Thanks for the insights @felixfontein.